Exa is an AI-native search engine and API that retrieves the most relevant web content for research and AI applications.
Exa (formerly Metaphor) built search for the AI era: neural models retrieve pages by meaning rather than keyword matching, making it strong for research questions where the right document is hard to name. Researchers use it to find companies, papers and experts; developers embed it as an API to ground LLM applications in live web data. Results can be filtered by category, recency and quality, and full-text retrieval returns clean content ready for analysis.
Exa offers trial credits for evaluation and usage-based pricing for the API; research use is supported through the web interface.
Researchers hunting hard-to-find sources and developers building AI applications that need reliable web retrieval.
Researchers find specialist sources keyword search misses, like technical blogs or niche company pages. AI developers use the API to give agents live, meaning-based retrieval instead of stale training data.
Run a natural-language search describing the kind of page you want, then refine with category and recency filters; developers can request API trial keys immediately.
Trial credits are available; sustained API usage is usage-priced.
Yes — the API is the core product, designed for retrieval in AI systems.
Versus Google or Bing APIs, Exa excels at semantic relevance and clean content extraction for AI use cases.
The index is refreshed continuously, making Exa suitable for monitoring and time-sensitive research questions.
Yes — structured discovery supports company and people searches with filters useful for market research and competitive analysis.
Neural retrieval ranks pages by meaning, surfacing high-quality niche content that keyword engines bury under optimized general results.
The API returns clean page content and links, designed to drop directly into retrieval pipelines for AI applications.
Yes — time filters prioritize fresh content, making Exa useful for monitoring markets and emerging research topics.
Yes — the API is designed first-class for retrieval inside AI products and agents.
Kimi is Moonshot AI's long-context assistant, built for reading entire books and document sets, with strong Chinese and English research support. OverviewKimi,…
Metaso is a Chinese AI search engine that delivers structured, ad-free research answers with outlines, mind maps, and full source citations. OverviewMetaso,…
Elicit is an AI research assistant that finds, summarizes, and extracts data from academic papers to accelerate literature reviews. OverviewElicit automates the…
